Semi-Clifford operations, structure of $\mathcal{C}_k$ hierarchy, and gate complexity for fault-tolerant quantum computation

Teleportation is a crucial element in fault-tolerant quantum computation and a complete understanding of its capacity is very important for the practical implementation of optimal fault-tolerant architectures. It is known that stabilizer codes support a natural set of gates that can be more easily implemented by teleportation than any other gates. These gates belong to the so called $\mathcal{C}_k$ hierarchy introduced by Gottesman and Chuang (Nature \textbf{402}, 390). Moreover, a subset of $\mathcal{C}_k$ gates, called semi-Clifford operations, can be implemented by an even simpler architecture than the traditional teleportation setup (Phys. Rev. \textbf{A62}, 052316). However, the precise set of gates in $\mathcal{C}_k$ remains unknown, even for a fixed number of qubits $n$, which prevents us from knowing exactly what teleportation is capable of. In this paper we study the structure of $\mathcal{C}_k$ in terms of semi-Clifford operations, which send by conjugation at least one maximal abelian subgroup of the $n$-qubit Pauli group into another one. We show that for $n=1,2$, all the $\mathcal{C}_k$ gates are semi-Clifford, which is also true for $\{n=3,k=3\}$. However, this is no longer true for $\{n>2,k>3\}$. To measure the capability of this teleportation primitive, we introduce a quantity called `teleportation depth', which characterizes how many teleportation steps are necessary, on average, to implement a given gate. We calculate upper bounds for teleportation depth by decomposing gates into both semi-Clifford $\mathcal{C}_k$ gates and those $\mathcal{C}_k$ gates beyond semi-Clifford operations, and compare their efficiency.